Recommendations as Treatments: Debiasing Learning and Evaluation

Tobias Schnabel, Adith Swaminathan, Ashudeep Singh, Navin Chandak, Thorsten Joachims
Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:1670-1679, 2016.

Abstract

Most data for evaluating and training recommender systems is subject to selection biases, either through self-selection by the users or through the actions of the recommendation system itself. In this paper, we provide a principled approach to handle selection biases by adapting models and estimation techniques from causal inference. The approach leads to unbiased performance estimators despite biased data, and to a matrix factorization method that provides substantially improved prediction performance on real-world data. We theoretically and empirically characterize the robustness of the approach, and find that it is highly practical and scalable.

Cite this Paper

BibTeX
@InProceedings{pmlr-v48-schnabel16, title = {Recommendations as Treatments: Debiasing Learning and Evaluation}, author = {Schnabel, Tobias and Swaminathan, Adith and Singh, Ashudeep and Chandak, Navin and Joachims, Thorsten}, booktitle = {Proceedings of The 33rd International Conference on Machine Learning}, pages = {1670--1679}, year = {2016}, editor = {Balcan, Maria Florina and Weinberger, Kilian Q.}, volume = {48}, series = {Proceedings of Machine Learning Research}, address = {New York, New York, USA}, month = {20--22 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v48/schnabel16.pdf}, url = {https://proceedings.mlr.press/v48/schnabel16.html}, abstract = {Most data for evaluating and training recommender systems is subject to selection biases, either through self-selection by the users or through the actions of the recommendation system itself. In this paper, we provide a principled approach to handle selection biases by adapting models and estimation techniques from causal inference. The approach leads to unbiased performance estimators despite biased data, and to a matrix factorization method that provides substantially improved prediction performance on real-world data. We theoretically and empirically characterize the robustness of the approach, and find that it is highly practical and scalable.} }
Endnote
%0 Conference Paper %T Recommendations as Treatments: Debiasing Learning and Evaluation %A Tobias Schnabel %A Adith Swaminathan %A Ashudeep Singh %A Navin Chandak %A Thorsten Joachims %B Proceedings of The 33rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2016 %E Maria Florina Balcan %E Kilian Q. Weinberger %F pmlr-v48-schnabel16 %I PMLR %P 1670--1679 %U https://proceedings.mlr.press/v48/schnabel16.html %V 48 %X Most data for evaluating and training recommender systems is subject to selection biases, either through self-selection by the users or through the actions of the recommendation system itself. In this paper, we provide a principled approach to handle selection biases by adapting models and estimation techniques from causal inference. The approach leads to unbiased performance estimators despite biased data, and to a matrix factorization method that provides substantially improved prediction performance on real-world data. We theoretically and empirically characterize the robustness of the approach, and find that it is highly practical and scalable.
RIS
TY - CPAPER TI - Recommendations as Treatments: Debiasing Learning and Evaluation AU - Tobias Schnabel AU - Adith Swaminathan AU - Ashudeep Singh AU - Navin Chandak AU - Thorsten Joachims BT - Proceedings of The 33rd International Conference on Machine Learning DA - 2016/06/11 ED - Maria Florina Balcan ED - Kilian Q. Weinberger ID - pmlr-v48-schnabel16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 48 SP - 1670 EP - 1679 L1 - http://proceedings.mlr.press/v48/schnabel16.pdf UR - https://proceedings.mlr.press/v48/schnabel16.html AB - Most data for evaluating and training recommender systems is subject to selection biases, either through self-selection by the users or through the actions of the recommendation system itself. In this paper, we provide a principled approach to handle selection biases by adapting models and estimation techniques from causal inference. The approach leads to unbiased performance estimators despite biased data, and to a matrix factorization method that provides substantially improved prediction performance on real-world data. We theoretically and empirically characterize the robustness of the approach, and find that it is highly practical and scalable. ER -
APA
Schnabel, T., Swaminathan, A., Singh, A., Chandak, N. & Joachims, T.. (2016). Recommendations as Treatments: Debiasing Learning and Evaluation. Proceedings of The 33rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 48:1670-1679 Available from https://proceedings.mlr.press/v48/schnabel16.html.

Related Material